NS 3.1 Changelog (Preliminary)

[Manta]

I think if I had a big icon over my head telling others to weld me, I wouldn't hide as much. At most, I'd step behind a pillar or wall when a fade blinks in.

[Spectre X]

at least, with the grenade thing, i can finally kill someone with a grenade  :p  instead of the usual: prime the grenade, get raped by a fade (or 2 or 3 skulks) and no one dies but me.


<edit> p.s. im used to cs nades, lol.

[Manta]

Eh, nades weren't horrible before. You just had to have some foresight; you couldn't pop out nades like you could a gun.

Just ask Goldfish.  

But I agree, nades will be more useful now. I still don't think you'll be able to use them like you could use, say, CS grenades. You'll have to plan a bit ahead, but you'll also be able to use them to react if the alien is far enough.

[That Annoying Kid]

How is this going to fix NS? I don't see anything about serverside performance!

[snapback]52623[/snapback]

^^

Touche indeed


*Tak glances at tanke momentarily
*Tak ponders asking on NSF

NS has always been not so hot on the serverside optimization, but still it would be nice to know whats going on, LB makes a good point

[edit]
yeah Manta, I've always wanted the hand grenades to have more bounce and oomph to the throw, so you can send them off a wall or two. I'm a fan of putting cooking in. NS has always made leaps and bounds in the small things that make it look nice on the HL engine, but that doesn't make up for the kind of server usage I see alot of people gripe about
[/edit]

[Scuzball]

Oh... mah god...

Skulks can launch of walls? How far? Does it play a sound? Can you combine it with a leap!? I wish I had more details!

I also wish I could play NS... right... now...

>_<

[tankefugl (in a tent)]

How is this going to fix NS? I don't see anything about serverside performance!

[snapback]52623[/snapback]

^^

Touche indeed


*Tak glances at tanke momentarily
*Tak ponders asking on NSF

NS has always been not so hot on the serverside optimization, but still it would be nice to know whats going on, LB makes a good point

I find it strange that most other server operators have said differently after 3.0.5, and even provided us with specific numbers. Best case so far has been Surftown, which had an average of 250% FPS increase for 24 players ... I don't really see the touché in LBs statement unless I get some numbers.

I'd love to get a hand on any performance statistics you got for LM, as well as the server configuration, list of mods etc., to see where the problem is.

[Goldy]

Eh, nades weren't horrible before. You just had to have some foresight; you couldn't pop out nades like you could a gun.

Just ask Goldfish. 

[snapback]52665[/snapback]

That was total luck, you said you didn't even know I was up there!

So what is the impulse for chuckle going to correspond with for marine? Taunt? That would make it TWICE as cool!

By the way, all of these requests don't get in the way of seeing things, do they? I've always thought there was a little too much "stuff" on my screen. Moving the text is helpful, but I hope the icons in the HUD and above players don't make for even more clutter.

[devicenull]

How is this going to fix NS? I don't see anything about serverside performance!

[snapback]52623[/snapback]

^^

Touche indeed


*Tak glances at tanke momentarily
*Tak ponders asking on NSF

NS has always been not so hot on the serverside optimization, but still it would be nice to know whats going on, LB makes a good point

I find it strange that most other server operators have said differently after 3.0.5, and even provided us with specific numbers. Best case so far has been Surftown, which had an average of 250% FPS increase for 24 players ... I don't really see the touché in LBs statement unless I get some numbers.

I'd love to get a hand on any performance statistics you got for LM, as well as the server configuration, list of mods etc., to see where the problem is.

[snapback]52674[/snapback]

http://www.lunixmonster.org/forums/index.php?showtopic=3392

Comparing the default server.cfg to LM's server.cfg.. the only thing I see that could effect FPS is
sv_voicecodec voice_speex
sv_voicequality 4

So I'm not sure whats causing it, unless its speex

Past the general hlds stuff you have to wait for LB :p

[aeroripper]

don't change voice quality 4

One thing i like about LM is its good voice clarity and is a lot nicer than most other servers...  

[devicenull]

don't change voice quality 4

One thing i like about LM is its good voice clarity and is a lot nicer than most other servers...

[snapback]52690[/snapback]

I didn't say we were chaging it, I said its the only non-default thing in the server.cfg that could change FPS/CPU usage

On a related note, I created a simple perl script that querys the server every minute to server CPU usage/FPS.
I'll eventaully set it up so that it reports the average CPU usage per player amount.

[tankefugl (in a tent)]

Awesome. You might want to measure server FPS at the same time, as the are related, but different metrics and more relevant to the entire issue with hitreg.

Edit: You may have 100% CPU usage in two cases, but 10 server FPS in one, and 100 in another, and the one with 100 will play smoothly.

[devicenull]

Awesome. You might want to measure server FPS at the same time, as the are related, but different metrics and more relevant to the entire issue with hitreg.

Edit: You may have 100% CPU usage in two cases, but 10 server FPS in one, and 100 in another, and the one with 100 will play smoothly.

[snapback]52695[/snapback]

It logs the output of the stats command.. which gives you
CPU   In    Out   Uptime  Users   FPS    Players
23.74 11.47 11.78    1155    84   85.34       5

[Bryan]

bows b4 devicenul.



Bryan

[Necrosis]

Patch sounds pretty awesome to me, randomised SG, cloaking tweaked, wallwalk tweaked up and a pounce-y  move thrown in.


Should be some good times to come!


As an evil constie I certainly look forward to trying to sneak my way into a stress test, failing that I think I'm going to clear my weekends to get some solid gaming in, this patch looks like it'll shake a few things up... or so I hope!!

[devicenull]

Leak me the beta :p

[Lightning Blue]

Awesome. You might want to measure server FPS at the same time, as the are related, but different metrics and more relevant to the entire issue with hitreg.

Edit: You may have 100% CPU usage in two cases, but 10 server FPS in one, and 100 in another, and the one with 100 will play smoothly.

[snapback]52695[/snapback]

Since I have other things running on the box, 100% CPU use with ANY FPS is unacceptable.

I don't really give a damn about good FPS, give me better CPU performance.

Caged at 8 players is doing 40% CPU, 90 FPS. I've never ever looked at FPS really.

She's a PIV 2.8 with 512 ram, win2k3 server, IDE.

 
win32 dlls/nsafk.dll
win32 dlls/unstuck_mm.dll
win32 dlls/logmod_mm_i386.dll


Very lean with plugins, dev could tell you what AMX has loaded.

[tankefugl (in a tent)]

Well then you're in a completely different domain.

HL attempts to squeeze as much as possible FPS from the CPU it can get. We can't reduce the CPU usage, but we can increase the FPS you get from the allowed CPU time with optimizations. I guess it's up to the OS or HL itself to reduce the allowed CPU usage, we can't controll it at all.

But if you find a way of restricting the allowed CPU usage, you should be able to restrict it down a bit and still have the same FPS as you had pre 3.0.5.

[Asal]

This is going to be fun.  

[Lightning Blue]

Nope, 2.01 used on average of .5%-10% of CPU for a full server on most maps. It was better then CS. Then Flay broke something and the CPU use went crazy like it's been for a very, very long time. We even tried figuring out what is was but other things came up.

I don't see really what the big deal is. It's not the allowed CPU use I am talking about, it's not using that much CPU in the first place.

[tankefugl (in a tent)]

We have control over the executed code per frame. The number of frames executed is for the HL code to decide. Even if every frame took one nanosecond, we would not be in control of the number of frames and could end up with 100% cpu.

Anyway, tried to tweak sys_tickrate?